Transcript Document

ADAPTIVE SYSTEMS &
USER MODELING
Alexandra I. Cristea
USI intensive course “Adaptive Systems” April-May 2003
Introduction
• Course site:
http://wwwis.win.tue.nl/~alex/HTML/USI/index.html
• Course schedule, principles, tasks, etc.
Module division
• I. Adaptive Systems and User
Modeling course
• II. Project work
Adaptive System course parts
1.
2.
3.
4.
Adaptive Systems, Generalities
User Modeling
Data representation for AS
Adaptive Systems, invited talk: Genetic
Algorithms
Project work parts
1.
2.
3.
4.
Presentation MOT
Presentation project assignments
Group work
Project and results presentation and
evaluation
Part 1: Adaptive Systems
Overview: AS
1.
2.
3.
4.
5.
6.
7.
8.
Adaptive Systems: Foundations
Artificial Adaptive Systems
Examples
General Classification
Applications
What can we adapt to?
Ultimate goal artificial AS?
Conclusion
Overview: AS
1.
2.
3.
4.
5.
6.
7.
8.
Adaptive Systems: Foundations
Artificial Adaptive Systems
Examples
General Classification
Applications
What can we adapt to?
Ultimate goal artificial AS?
Conclusion
Foundations of Adaptive
Computation:
Natural Adaptive Systems
What are Adaptive Systems in
Nature?
Examples?
Natural Systems
• How do adaptive systems in nature
compute?
• (De-)centralized/collective computation
• Computation over spatial extent
• Probabilistic computation
• Computation in continuous-state systems
• Computation in neural systems
Overview: AS
1.
2.
3.
4.
5.
6.
7.
8.
Adaptive Systems: Foundations
Artificial Adaptive Systems
Examples
General Classification
Applications
What can we adapt to?
Ultimate goal artificial AS?
Conclusion
Artificial Adaptive Systems
Types of Artificial
Adaptive Systems
• Adaptive Hypermedia, Agents, Game of Life,
Ant Algorithms, Genetic Algorithms, Artificial
Life, Genetic Art, Brain Building, Genetic
Programming, Cellular Automata, Cellular
Computing, Cellular Neural Networks, Cellular
Programming, Complex Adaptive Systems,
Quantum Computing, Cybernetics, Reversible
Computing, DNA Computing, Self-Replication,
Evolutionary Computation, Evolvable Hardware,
Virtual Creatures, Flocking Behaviour, etc.
Overview: AS
1.
2.
3.
4.
5.
6.
7.
8.
Adaptive Systems: Foundations
Artificial Adaptive Systems
Examples
General Classification
Applications
What can we adapt to?
Ultimate goal artificial AS?
Conclusion
Artificial Adaptive Systems
Examples
Example1
Evolving artificial creatures, Karl Sims:
http://biota.org/ksims/blockies/index.html#video
Example2
• Ants
TSP pb.
Ex.3: NN: spatial forms
Ex. 4: NN:OCR
Ex.5: intelligent agent Steve
http://www.isi.edu/isd/VET/steve-demo.html
Overview: AS
1.
2.
3.
4.
5.
6.
7.
8.
Adaptive Systems: Foundations
Artificial Adaptive Systems
Examples
General Classification
Applications
What can we adapt to?
Ultimate goal artificial AS?
Conclusion
General Classification of AS
• Software
• Hardware
• Combined
Example: combined
• Khepera robot
Elements Technical Information
Processor
Motorola 68331, 25MHz [improved]
RAM
512 Kbytes [improved]
Flash
Motion
512 Kbytes
Programmable via serial port [new]
2 DC brushed servo motors with incremental encoders
Speed
Max: 60 cm/s, Min: 2 cm/s
Sensors
8 Infra-red proximity and ambient light sensors with up to 100mm range
I/O
3 Analog Inputs (0-4.3V, 8bit)
Power
Power Adapter
Rechargeable NiMH Batteries[improved]
1 hour, moving continuously [improved].
Autonomy
Communica
tion
Extension
Standard Serial Port, up to 115kbps [improved]
Size
Diameter: 70 mm
Height: 30 mm
Approx 80 g
Weight
Expansion modules can be added to the robot
Overview: AS
1.
2.
3.
4.
5.
6.
7.
8.
Adaptive Systems: Foundations
Artificial Adaptive Systems
Examples
General Classification
Applications
What can we adapt to?
Ultimate goal artificial AS?
Conclusion
Applications of Artificial
Adaptive Systems
Applications of Adaptive
Systems
• expert systems
– (e.g. medical diagnosis)
• data mining
– (e.g. search engines)
• computational linguistics
• games
More Applications of Adaptive
Computation
• Parallel computing:
– evolution of cellular automata
• Molecular biology:
– molecular evolution, design of useful molecules,
protein design
• Computer security:
– immune systems for computers
• Intelligent agents and robotics
• Scientific modeling:
– evolution, ecologies, economies, insect societies,
immune systems, organizations
Overview: AS
1.
2.
3.
4.
5.
6.
7.
8.
Adaptive Systems: Foundations
Artificial Adaptive Systems
Examples
General Classification
Applications
What can we adapt to?
Ultimate goal artificial AS?
Conclusion
What can we adapt to?
• What kind of information can we use to
adapt, in general?
• From whom/ what do we get this
information?
• What means adaptation in this context?
What can we adapt to?
• What kind of information can we use to
adapt, in general?
– External:
• Static Variables values: Light intensity,
• Dynamics: Changes,
• Other participants’ behavior
– Internal:
• Needs: hunger
– Prediction: (anticipation)
What can we adapt to?
• From whom/ what do we get this
information?
– Other participants
– Existing variables
What can we adapt to?
• What means adaptation in this context?
– The adaptive system reacts to the
environment (static, dynamics) and to itself
towards some benefit
Overview: AS
1.
2.
3.
4.
5.
6.
7.
8.
Adaptive Systems: Foundations
Artificial Adaptive Systems
Examples
General Classification
Applications
What can we adapt to?
Ultimate goal artificial AS?
Conclusion
A Comparison between
Adaptive and Adaptable Systems
Gerhard Fischer 1 HFA Lecture, OZCHI’2000
Ultimate Goal of Artificial
Adaptive Systems?
Intelligence
Conclusions
• Man is trying to imitate nature with
artificial AS
• Why?
• Because man-made machines with
predefined behavior cannot cover all
aspects
• Note:
Adaptation < Learning < Intelligence
Conclusions 2
• Adaptation in general doesn’t mean to a
human […]
• However, adaptation to a human is more
challenging!